Method of detecting yarn unevenness

Abstract

Method of detecting yarn unevenness appearing as periodic variations of a spun yarn. First, plural yarn samples are taken. Then, k distinct spectrograms are obtained according to the number of the samples. These k spectrums are averaged to create an overall spectrogram. The height of adjacent channels in the overall spectrogram are averaged to create a reference spectrogram. Either the overall spectrogram of the distinct spectrograms are compared with the reference spectrogram of find those portions of the channels that exceed the reference spectrograms in height. These exceeding portions are displayed. Therefore, realistic defective added nonuniformities contained in the spun yarn can be easily detected.

What is claimed is: 
     
       1. A method of detecting uneven yarn components by taking plural samples from a yarn to be investigated, moving the samples, extracting signals proportional to variations of weights of the samples, analyzing the signals in terms of frequency or wavelength to produce signal components of different wavelengths indicative of any uneven yarn component, classifying the signal components into plural channels, and creating distinct spectrograms representing rates of variation of the signal components in terms of height of channel, said method comprising the steps: obtaining k distinct spectrograms CH n  (n=1, 2, . . . , k) according to the number of said yarn samples;   averaging said k distinct spectrograms with respect to each signal component in each channel thereof to create an overall spectrogram CH OA  as one diagram which may contain defective unevenness common to all the samples;   averaging said overall spectrogram CH OA  in a manner averaging heights between adjacent channels thereof to create a realistic reference spectrogram CH RF  in which periodic defects have been removed from said overall spectrogram CH OA  ;   comparing said reference spectrogram CH RF  and said overall spectrogram CH OA  to find exceeding portions of the channels in the overall spectrogram CH OA , said exceeding portions exceeding said reference spectrogram CH RF  in height; and   displaying these exceeding portions as an average added unevenness spectrogram CH AD .   
     
     
       2. The method of claim 1, wherein said exceeding portions of said exceeding channels are displayed in colors differentiated according to its amount of excess. 
     
     
       3. A method of detecting uneven yarn components by taking plural samples from a yarn to be investigated, moving the samples, extracting signals proportional to variations of weights of the samples, analyzing the signals in terms of frequency or wavelengths to produce signal components of different wavelengths indicative of any uneven yarn component, classifying the signal components into plural channels, and creating distinct spectrograms representing rates of variation of the signal components in terms of height of channel, said method comprising the steps: obtaining k distinct spectrograms CH n  (n=1, 2, . . . , k) according to the number of said yarn samples;   averaging said k distinct spectrograms with respect to each signal component in each channel thereof to create an overall spectrogram CH OA  as one diagram which may contain defective unevenness common to all the samples;   averaging said overall spectrogram CH OA  in a manner averaging heights between adjacent channels thereof to create a realistic reference spectrogram CH RF  by removing periodic defects from said overall spectrogram CH OA  ;   comparing said distinct spectrograms Ch n  and said reference spectrogram CH RF  to find exceeding portions of the channels in the distinct spectrograms, said exceeding portions exceeding said reference spectrogram CH RF  in height; and   displaying these exceeding portions as an average added unevenness spectrogram CH AD .   
     
     
       4. The method of claim 3, wherein said exceeding portions of said exceeding channels are displayed in colors differentiated according to its amount of excess.